fix: sort signed URL replacements by position

api/models/dataset.py

@@ -8,6 +8,7 @@ import os
 import pickle
 import re
 import time
+from operator import itemgetter
 from datetime import datetime
 from json import JSONDecodeError
 from typing import Any, cast
@@ -867,6 +868,7 @@ class DocumentSegment(Base):
             signed_urls.append((match.start(), match.end(), signed_url))
 
         # Reconstruct the text with signed URLs
+        signed_urls.sort(key=itemgetter(0))
         offset = 0
         for start, end, signed_url in signed_urls:
             text = text[: start + offset] + signed_url + text[end + offset :]

api/tests/unit_tests/models/test_dataset_models.py

@@ -653,6 +653,43 @@ class TestDocumentSegmentIndexing:
         assert f"/files/{denied_upload_file_id}/file-preview?timestamp=" not in signed
         assert f"/files/{denied_upload_file_id}/file-preview)" in signed
 
+    def test_document_segment_sign_content_handles_mixed_preview_order(self):
+        """Test that sign_content preserves content when file-preview appears before image-preview."""
+        # Arrange
+        file_preview_id = "1602650a-4fe4-423c-85a2-af76c083e3c4"
+        image_preview_id = "e2a4f7b1-1234-5678-9abc-def012345678"
+        segment = DocumentSegment(
+            tenant_id=str(uuid4()),
+            dataset_id=str(uuid4()),
+            document_id=str(uuid4()),
+            position=1,
+            content=(
+                f"file-first: ![file](/files/{file_preview_id}/file-preview) "
+                f"then-image: ![image](/files/{image_preview_id}/image-preview)"
+            ),
+            word_count=1,
+            tokens=1,
+            created_by=str(uuid4()),
+        )
+        mock_scalars_result = self._mock_scalars_result([file_preview_id, image_preview_id])
+
+        # Act
+        with (
+            patch.object(dataset_module.dify_config, "SECRET_KEY", "secret", create=True),
+            patch("models.dataset.db.session.scalars", return_value=mock_scalars_result),
+            patch("models.dataset.time.time", return_value=1700000000),
+            patch("models.dataset.os.urandom", return_value=b"\x00" * 16),
+        ):
+            signed = segment.get_sign_content()
+
+        # Assert
+        file_signed = f"/files/{file_preview_id}/file-preview?timestamp="
+        image_signed = f"/files/{image_preview_id}/image-preview?timestamp="
+        assert file_signed in signed
+        assert image_signed in signed
+        assert signed.index(file_signed) < signed.index(image_signed)
+        assert signed.count("&sign=") == 2
+
     def test_document_segment_with_answer_field(self):
         """Test creating a document segment with answer field for QA model."""
         # Arrange